We study the mechanism and regulation of protein synthesis in eukaryotic cells focusing on regulation by GTP-binding proteins and protein phosphorylation. The first step of protein synthesis is binding the initiator Met-tRNA to the small ribosomal subunit by the factor eIF2. The eIF2 is a GTP-binding protein and during the course of translation initiation the GTP is hydrolyzed to GDP. The eIF2 is released from the ribosome in complex with GDP and requires the guanine-nucleotide exchange factor eIF2B to convert eIF2-GDP to eIF2-GTP. This exchange reaction is regulated by a family of stress-responsive protein kinases that specifically phosphorylate the alpha subunit of eIF2 on serine at residue 51, and thereby covert eIF2 into an inhibitor of eIF2B. Among the family of eIF2alpha kinases are GCN2, which is activated under conditions of amino acid starvation, and PKR, which is activated by double-stranded RNA and downregulates protein synthesis in virally infected cells. Truncation analyses revealed that eIF2alpha residues 1-180 form the minimal substrate for efficient phosphorylation of serine-51 by GCN2 and PKR both in vitro and in vivo. Site-directed and random mutational analyses identified a large number of mutations throughout the N-terminal OB-fold domain (residues 1-90) of yeast eIF2alpha that impaired translational regulation (reference 2). Any mutation at glutamate-49 or the remote aspartate-83 blocked translational regulation, however only a subset of the mutations impaired serine-51 phosphorylation. Substitution of alanine for aspartate-83 eliminated serine-51 phosphorylation both in vivo and in vitro establishing the importance of residues remote from the phosphorylation site for kinase substrate recognition. Other mutations that blocked translational regulation, but not serine-51 phosphorylation, impaired the binding of eIF2B to phosphorylated eIF2alpha. Thus, the eIF2alpha kinases and eIF2B recognize the same surface and overlapping determinants on eIF2alpha. In collaboration with Frank Sicheri we determined the structure of the PKR kinase domain in complex with eIF2alpha. This structural analysis revealed that eIF2alpha binds to the C-terminal lobe making intimate contact with helix alphaG, while catalytic domain dimerization is mediated by a back-to-back orientation of the kinase N-terminal lobes (reference 4). Positioning of the eIF2alpha aspartate-83 residue near PKR helix alphaG places the serine-51 residue near the active site of the kinase. Consistent with the structural data, mutations in PKR helix alphaG specifically impair phosphorylation of eIF2alpha. Moreover, mutations that activate PKR map to the catalytic domain dimer interface and promote kinase domain dimerization. Conversely, mutations that disrupt a conserved salt-bridge in the dimer interface block PKR autophosphorylation and eIF2alpha phosphorylation. Finally, mutation of the conserved threonine-446 autophosphorylation site in PKR impairs eIF2alpha phosphorylation and viral pseudosubstrate binding. We propose an ordered mechanism of PKR activation in which catalytic domain dimerization triggers autophosphorylation and specific substrate recognition (reference 5). The GTP-binding protein eIF5B catalyzes ribosomal subunit joining in the final step of translation initiation. The eIF5B is an ortholog of prokaryotic translation initiation factor IF2. Previous studies revealed that eIF5B consists of four domains that structurally assemble to form a chalice-shaped molecule. The G domain plus domains II and III form the cup of the chalice, a long alpha helix forms the stem, and domain IV is the base of the chalice. In addition, we previously showed that the domain IV of eIF5B binds to the C-terminal tail of the factor eIF1A (an ortholog of prokaryotic factor IF1). We propose that the eIF5B-eIF1A interaction is important for binding eIF5B to the ribosome and possibly also for release of both factors from the ribosome following subunit joining and GTP hydrolysis by eIF5B. The G domain of eIF5B contains the hallmarks associated with typical GTP-binding proteins including the conserved switch 1 and switch 2 motifs. Mutation of the conserved threonine residue in switch 1 abolished GTP hydrolysis, but did not impair subunit joining in vitro. Intragenic suppressors of the switch 1 mutation uncoupled eIF5B GTPase and translational stimulatory activities indicating a regulatory rather than mechanical role for eIF5B GTP hydrolysis in translation initiation. We propose that in the presence of GTP eIF5B binds the ribosome and promotes subunit joining, which in turn triggers GTP hydrolysis leading to the factor's release from the ribosome. Mutation of the conserved glycine in switch 2 of eIF5B impaired GTP binding, GTP hydrolysis, translation initiation and yeast cell growth. Intragenic suppressors of the slow-growth phenotype associated with the switch 2 mutation mapped to switch 1 and to helix 8 (linking domains II and III). The intragenic suppressor in switch 1 restored both the GTP binding and GTPase activities of eIF5B revealing that the universally conserved glycine in switch 2 is not absolutely essential. Interestingly, the intragenic suppressors in switch 1 and helix 8 are located close to contact sites with switch 2, and the suppressor mutations are predicted to allosterically affect the position of switch 2. We propose that mutation of the conserved glycine in switch 2 alters the structure of the eIF5B active site, and that the two intragenic suppressor mutations restore a favorable geometry to the eIF5B active site by re-positioning switch 2 into a preferred location. As the switch 2 mutation and the switch 1 suppressor mutation map to elements conserved in all GTP-binding proteins, we believe that this interaction may be of importance for all GTP-binding proteins.